Re: [PATCH 5/9] perf, bpf: save bpf_prog_info in a rbtree in perf_env

[Song Liu]

> On Feb 14, 2019, at 4:26 AM, Jiri Olsa <jo...@redhat.com> wrote:
> 
> On Fri, Feb 08, 2019 at 05:17:01PM -0800, Song Liu wrote:
> 
> SNIP
> 
>> diff --git a/tools/perf/util/env.h b/tools/perf/util/env.h
>> index d01b8355f4ca..5894a177b7cf 100644
>> --- a/tools/perf/util/env.h
>> +++ b/tools/perf/util/env.h
>> @@ -3,7 +3,10 @@
>> #define __PERF_ENV_H
>> 
>> #include <linux/types.h>
>> +#include <linux/rbtree.h>
>> #include "cpumap.h"
>> +#include "rwsem.h"
>> +#include "bpf-event.h"
>> 
>> struct cpu_topology_map {
>>      int     socket_id;
>> @@ -64,6 +67,8 @@ struct perf_env {
>>      struct memory_node      *memory_nodes;
>>      unsigned long long       memory_bsize;
>>      u64                     clockid_res_ns;
>> +    struct rw_semaphore     bpf_info_lock;
> 
> why's the lock needed?
> 
> jirka

It protects the retries for bpf_prog_info and btf. For perf-top, 
we will have one thread writing to the trees, while the main 
thread reading from them. 

Let me add comments to clarify. 

Thanks,
Song